See how it works (in Russian) A one-minute walkthrough: what a parent writes in WhatsApp and how the assistant answers from the school’s base.
Free website check (in Russian) In a minute we show whether a parent finds the price, places and hours on your site — or messages you.
What this is
There is no “School” template in this build: it needs a class-schedule module, which is not built yet. A training centre runs on the shared modules configured for it. Not in this build:
- Class schedule, attendance and teacher substitution — no module.
- Hours saved and parent satisfaction: not measured, as there were no live users or surveys.
- Mailings to parents: the platform never writes first.
Modules and AI employees
AI employees with work here: support, knowledge, sales, finance and BI analyst. The modules used in the internal check, each switched on separately:
- Knowledge base: confirmed answers on prices, schedule, enrolment, refunds and discounts.
- Support: parents’ requests with a queue, a deadline and handover to a person.
- Sales: website requests as leads, enrolment by stage.
- Finance: the enrolment budget, invoices for approval.
- Analytics: monthly figures with a source for every number.
More detail
Same in every industry
Company isolation in the database, a closed permission register, a decision journal the application cannot change or erase, and two-person sign-off where needed. AI employees only prepare drafts; payments, signatures and decisions about people stay with staff. A customer gets a staff member’s reply or, in the website chat when it is on, the verbatim text of a confirmed answer.
What this rests on
An automated internal check of the product, «AAA AI School», on made-up training-centre data — not a customer result. There have been no customer pilots yet, and we do not present an internal run as anyone’s result.
More detail
The run: 46 knowledge-base answers, all confirmed by a second person; of 13 parents’ questions, 10 were answered from the base with sources and 3 outside it got an honest “no data”; no question was lost. All fifteen AI employees were set up, one was moved into work by two people. Only built-in rules ran, with zero external calls: a product acceptance test, not a measure of a model.
A question in Kazakh against a Russian base got “no data”: confirmed answers are needed in each language.
Basis: the report of the automated education-centre scenario run; the registry of modules and industry templates. This page describes what exists in this build, not what is planned.